A univariate time series model, in the most general form given by
where Δd yt is the dth difference of yt. This is a generalization of the autoregressive moving average (ARMA (p, q)) model used to describe a non-stationary process that becomes stationary after being differenced d times. When d is a fraction the process is sometimes referred to as an autoregressive fractionally integrated moving average, or ARFIMA process.